DE602004008738T2 - GENERAL TWO-STAGE DATA ASSESSMENT - Google Patents

Abstract

A wireless transmit/receive unit (WTRU) is configure to receive and sample wireless signals in a shared spectrum where the wireless signal comprise encoded symbols. The WTRU has a channel estimation device configured to process received signal samples to produce an estimate of a channel response of the received signals corresponding to a matrix H. The channel estimation device is preferably configured to process the received signal samples to produce an estimate of noise variance of the received signals. The WTRU preferably has a two stage data estimator that includes a channel equalizer and a despreader. The channel equalizer is configured to process received signal samples using the estimated channel response matrix H and the estimate of noise variance to produce a spread signal estimate of the received signals. The despreader is configured to process the spread signal estimate of the received signals produced by said channel equalizer to recover encoded symbols of the received signals.

Description

Field of the invention

The The present invention relates to wireless communication systems. In particular, the present invention is based on the data estimation in aligned with such systems.

background

In Wireless systems use joint-detection (JD) to detect intersymbol interference (ISI) and multiple access interference (MAI). JD is characterized by good performance but high complexity. Even at Using Cholesky approximation or block Fourier transforms with Cholesky decomposition algorithms is the complexity the JD is still very high. When used in a wireless receiver JD becomes, prevents their complexity, that the receiver implemented efficiently. This proves the need for alternative Algorithms that are not only simple in implementation, but also have a good performance.

Around overcome this problem became recipients of the prior art based on a channel equalizer followed developed by a code spreader. These types of receivers are referred to as Single User Detection (SUD) receivers, because the detection method in contrast to JD receivers not requires the knowledge of channel sharing codes of other users. The SUD tends to for most interesting data rates do not have the same performance how to provide the JD, even if its complexity is very low is.

A solution for the The above problem is known, for example, from Vollmer et al. "Joint-Detection using Fast Fourier Transforms in TD-CDMA based Mobile Radio Systems ", Proceedings of the International Conference at Telecomm., ICT, 1999, pages 1-7, XP002190679. This prior art shows that the joint detection method is equivalent to the solution a least squares problem with a Toeplitz block system matrix is. By extending this matrix to a block-circulating matrix Is it possible, to diagonalize the matrix with fast Fourier transforms.

consequently There is a need for low complexity and high density data detectors Power.

Summary

symbols have to be recovered from signals that are in a shared Spectrum are received. Codes in the shared Spectrum received signals are using a block Fourier transform (FT), creating a block-diagonal code matrix. A channel response of the received signals is estimated. The Channel response is extended and modified to a blockcirculatory Matrix to generate, and a block FT is performed, wherein a block diagonal channel response matrix is generated. The block diagonal Code matrix is combined with the block diagonal channel response matrix. The received signals are sampled and using the combined block diagonal code matrix and the block diagonal Channel response matrix processed with a Cholesky algorithm. For a result of the Cholesky algorithm a block inverse FT is performed to spread symbols to create. The splayed symbols are despread to symbols recover the received signals.

Brief description of the drawings

1 Fig. 10 is a block diagram showing two-stage data detection.

2 Fig. 10 is a block diagram of one embodiment of two-stage data detection.

3 Figure 10 is a block diagram of the code assignment to reduce the complexity of two-level data detection.

4A - 4D are block diagrams of the use of lookup tables to determine Λ _R.

Detailed description of the preferred Embodiment (s)

The The present invention will be described with reference to the drawings described, wherein like numbers indicate the same elements throughout.

One two-stage data estimator Can be used in a wireless transmitter / receiver unit (WTRU) or base station are used when all communications to be detected by the estimator a similar Learn channel response. Although the following in connection with the preferred from the third generation partnership project (3GPP) proposed wideband code division multiple access (W-CDMA) communication system is described, it is applicable to other systems.

1 Figure 4 is a simplified block diagram of a receiver using a two-stage data estimator 55 , An antenna 50 or an antenna arrangement receives radio frequency signals. The signals are from a scanning device 51 typically sampled at the chip rate or a multiple of the chip rate, producing a receive vector r . A channel estimator 53 , which uses a reference signal such as a midamble sequence or a pilot code, estimates the channel response for the received signals as a channel response matrix H. The channel estimator 53 also estimates the noise spread σ ² .

The channel equalizer 52 takes the receive vector r and equalizes it using the channel response matrix H and the noise spread σ ² , generating a spread symbol vector s . Using codes C of the receiving signals, a despreader despreads 54 the spread symbol vector s , where the estimated symbols d are generated.

oderWith Joint Detection (JD), a formula for the minimum mean square error (MMSE) with respect to the symbol vector d can be expressed as:

or

ist die Schätzung von d, r ist der Empfangssignalvektor, A ist die Systemmatrix, R_n ist die Kovarianzmatrix der Rauschfolge, R_d ist die Kovarianzmatrix der Symbolfolge, und der Begriff (·)^H bezeichnet die komplex konjugierte Transformations-(Hermitesche)Operation. Die Ausmaße und Strukturen der obigen Vektoren und Matrizen hängen von der spezifischen Systemkonstruktion ab. Normalerweise haben unterschiedliche Systeme unterschiedliche Systemparameter, wie etwa die Rahmenstruktur, die Datenfeldlänge und die Länge der Verzögerungsspreizung.

is the estimate of d , r is the received signal vector , A is the system matrix, R _n is the covariance matrix of the noise sequence, R _d is the covariance matrix of the symbol sequence, and the term (·) ^H denotes the complex conjugate transform (Hermitian) operation. The dimensions and structures of the above vectors and matrices depend on the specific system design. Typically, different systems have different system parameters, such as the frame structure, the data field length, and the delay spread length.

The Matrix A has for different systems different extent values, and the dimensions of the matrix A hang from the data field length, the number of codes, the spreading factor and the length of the delay spread from. By way of example, the matrix A has for the transmission of 8 codes respectively with the spreading factor 16 dimensions from 1032 times 488 for a WCDMA TDD system, when the burst type 1 is used and the delay spread is 57 chips long is while the matrix A for a TD-SCDMA system for a delay spread of 16 chips in length dimensions of 367 times 176.

oder

White noise and uncorrelated symbols with unit energy are assumed to be R _n = σ ² I and R _d = I, where I denotes the identity matrix. Substituting these in Equation 1 and 2 gives:

or

The received signal can be viewed as a composite signal denoted s , which has passed through a single channel. The receive signal r can be represented by r = H s , where H is the channel response matrix and s is the composite spread signal. H takes the following form:

In Equation (5), W is the length of the channel response and is therefore equal to the length of the delay spread. Typically W = 57 for the W-CDMA Time Division Duplex (TDD) burst type 1 and W = 16 for Time Division Multiplexed CDMA (TD-CDMA). The composite spread signal s can be expressed as s = C d , where the symbol vector is d : d = (i 1 , d 2 , ..., d KN ) T Equation (6) and the code matrix C is: C = [C (1) , C (2) , ..., C (K) ] Equation (7) With:

Q, K and N _S respectively denote the spreading factor (SF), the number of active codes and the number of symbols carried on each channel division code. c (K) i is the ith element of the kth code. The matrix C is a matrix of size N _s * Q times N _s * K.

R_C = CC^H. Wenn ŝ das geschätzte gespreizte Signal bezeichnet, kann die Gleichung (9) in zwei Stufen ausgedrückt werden:Substituting A = HC into equation (4) yields:

R _C = CC ^H. If ŝ denotes the estimated spread signal, equation (9) can be expressed in two stages:

Step 1:

• ŝ  = H H (MR C H H  + σ 2 I) -1 r  Equation (10)

Level 2:

The first stage is the generalized channel equalization stage. It estimates the spread signal s by an equalization method according to Equation 10. The second stage is the despreading stage. The symbol sequence d is recovered by a despreading method according to Equation 11.

The matrix R _C in equation 9 is a block diagonal matrix of the form:

The block R ₀ in the diagonal is a square matrix of size Q. The matrix R _C is a square matrix of size N _S · Q.

Since the matrix R _{C is} a block-circulant matrix, the fast block Fourier transform (FFT) can be used to implement the algorithm. With this approach, the matrix R _{C can be} decomposed as: R C = F -1 (Q) Λ R F (Q) Equation (13) With F (Q) = F ns ⊗ I Q Equation (14)

F _Ns is the N _S point FFT matrix, I _Q is the unit matrix of size Q and the notation ⊗ is the Kronecker product. By definition, the Kronecker product Z is the matrix X and Y (Z = X ⊗ Y):

x _{m, n} is the (m, n) th element of the matrix X. For each F _(Q) , an Ns-point FFT is performed Q times. Λ _R is a block diagonal matrix whose diagonal blocks are:
F _(Q) R _C (:, 1: Q). This means, diag (Λ R ) = F (Q) R C (:, 1: Q), equation (16) R _C (:, 1: Q) denotes the first Q columns of the matrix R _C.

wobei H_i, i = 0, 1, ..., L-1 jeweils eine quadratische Matrix der Größe Q ist. L ist die Anzahl von Datensymbolen, die von der Verzögerungsspreizung des Ausbreitungskanals beeinflußt ist und wird ausgedrückt als:

The block-circulating matrix can be decomposed into simple and efficient FFT components, making a matrix more efficient and less complex. Normally, the large matrix inverse is more efficient when executed in the frequency domain rather than the time domain. For this reason, it is advantageous to use FFT, and the use of a block circular matrix enables the efficient FFT implementation. With the proper partition, the matrix H can be expressed as an approximately block-circular matrix of the following form:

where H _i, i = 0, 1, ..., L-1 is in each case a square matrix of size Q. L is the number of data symbols affected by the delay spread of the propagation channel and is expressed as:

To enable the block FFT decomposition, H can be expanded and modified into an exactly block-circulant matrix of the following form:

The block circulatory matrix H _C is obtained by extending the columns of the matrix H in the equation (17) by sequentially shifting one element block down in turn.

The matrix H _C can be decomposed by block FFT as: H C = F -1 (Q) Λ H F (Q) Equation (20)

Λ _H is a block diagonal matrix whose diagonal blocks are:
F _(Q) H _C (:, 1: Q), since diag (Λ H ) = F (Q) H C (:, 1: Q) equation (21)

H _C (:, 1: Q) denotes the first Q columns of the matrix H _C. From equation (20) can H H C be defined as: H H C = F -1 (Q) Λ H H F (Q) Equation (22)

By replacing the matrices R _C and H _C in Equation 10, ŝ is obtained: ŝ = F -1 (Q) Λ H H (Λ H Λ R Λ H H + σ 2 I) -1 F (Q) r Equation (23)

For a zero enforcing (IF) solution, equation 19 is simplified to: ŝ = F -1 (Q) Λ -1 R Λ -1 H F (Q) r Equation (24)

The Matrix inverses in equations (23) and (24) can be found under Using the Cholesky decomposition and forward and backward substitutions.

undIn a special case of K = SF, (where the number of active codes is equal to the spreading factor), the matrix R _C becomes a scalar-diagonal matrix with identical diagonal elements equal to SF. In this case, equations (10) and (11) are reduced to:

and

Equation (25) can also be expressed in the following form:

undWith FFT, Equations (25) and (27) can each be realized as:

and

Λ _H is a diagonal matrix whose diagonal is F · H (:, 1), where H (:, 1) denotes the first column of the matrix H. The term (·) * denotes the conjugation operator.

2 is a preferred block diagram of the channel equalizer 15. A code matrix C is input to the channel equalizer 15. A Hermite device 30 takes a complex conjugate transpose of the code matrix C, C ^H. The code matrix C and its hermit are from a multiplier 32 multiplied, where CC ^{H is} generated. A block FT performed for CC ^H generates the block diagonal matrix Λ _R.

The channel response matrix H is from an extension and modification device 36 extended and modified, where H _{C is} generated. A block FT 38 takes H _C and creates a block diagonal matrix Λ _H. A multiplier 40 multiply Λ _H and Λ _R together, producing Λ _H Λ _R. A Hermite device 42 takes the complex conjugate transpose of Λ _H , what Λ H H generated. A multiplier 44 multiplied Λ H H with Λ _H Λ _R , what Λ H Λ R Λ H H generated, and an adder 46 adds σ ² I, which Λ H Λ R Λ H H + σ 2 I generated.

A Cholesky decomposition device 48 creates a Cholesky factor. A block FT 20 takes a block FT of the receive vector r . Using the Cholesky Factor and the FT of r are from a forward substitution device 22 and a backward substitution device performs forward and backward substitution.

A conjugation device 56 takes the conjugate of Λ _H , where Λ ^* _{H is} generated. The result of the backward substitution is on a multiplier 58 multiplied by Λ ^* _H. A block inverse FT 60 takes a block inverse FT of the multiplied result, producing ŝ .

According to one another embodiment The present invention provides an approximate solution in which a generalized two-stage data detection method block diagonal approximation is. The block diagonal approximation comprises in the approximation process Posts outside the diagonal as well as diagonal entries.

wobei als c bezeichnete Elemente Konstanten darstellen und immer gleich der Anzahl der Kanalteilungscodes sind, d.h. c = K. Die mit x bezeichneten Elemente stellen einige Variablen dar, deren Werte und Stellen sich mit verschiedenen Kombinationen von Kanalteilungscodes ändern. Ihre Stellen ändern sich abhängig von der Kombination von Codes gewissen Mustern folgend. Als ein Ergebnis sind nur ein paar von ihnen ungleich null. Wenn die Codeleistung berücksichtigt wird und nicht die Einheitsleistung ist, ist das Element c gleich der Gesamtleistung der übertragenen Codes. Eine gute Näherung für die Matrix R₀ ist, den konstanten Teil aufzunehmen und den variablen Teil zu ignorieren als:

As an example, consider the case of four channel division codes. R ₀ , a combination of four channel division codes, has a constant block diagonal part that does not change with the different combinations of codes, and a edge part that changes with the combinations. In general, R _{0 has} the following structure:

where elements denoted as c represent constants and are always equal to the number of channelization codes, ie, c = K. The elements labeled x represent some variables whose values and locations change with different combinations of channelization codes. Their digits change depending on the combination of codes following certain patterns. As a result, only a few of them are non-zero. If the code power is taken into account and not the unit power, the element c is equal to the total power of the transmitted codes. A good approximation for the matrix R ₀ is to pick up the constant part and ignore the variable part as:

wobei P_Mittel die mittlere Codeleistung ist, die erhalten wird durch

In this case, the approximation R ^ _{0 contains} only a constant part. Regardless of which codes are transmitted, R ^ ₀ depends only on the number of active codes, and R ^ _C can be decomposed as shown in equation (13). The block diagonal of Λ _R or F _(Q) R 1 _C (:, 1: Q) can be pre-calculated using a FFT for different numbers of codes and stored as a look-up table. This reduces the computational complexity by not calculating F _(Q) R _C (:, 1: Q). In the case where the code power is considered and not the unit power, the element c becomes the total power of active codes (ie, c = P _T , where P _{T is} the total power of active codes). The matrix R ^ ₀ can be expressed as:

where P _{means is} the average code power obtained by

In this case, a scaling factor P _{average should} be applied in the method.

Other variants of the block diagonal approximation method can be derived by including more entries than the constant block diagonal part. This improves performance, but entails more complexity because now, by including variable entries, the FFT must be recalculated for F _(Q) R _C (:, 1: Q) as required when the codes change. Using more entries enhances the exact solution because all of the non-diagonal entries are included for processing.

at Given a given number of channel division codes, one can use the code sets for different ones Derive combinations of channelization codes that have a common have constant part of the correlation matrix whose values are equal the number of channel division codes or the total power of the channel division codes are if the code does not have the unit code performance. To the implementation with low complexity can facilitate the assignment of channelization codes or resource units According to the rules, make a code set random the code sets singled out, who have a common constant part, and these codes are assigned in the extracted code set. When assigning four codes, for example, the code sets [1, 2, 3, 4], [5, 6, 7, 8], [9, 10, 11, 12], ... a common constant Part in their correlation matrix. If the channel allocation of four Codes should be one of these code sets for the optimum Computational efficiency can be used.

3 Fig. 10 is a flow chart of such a channel code assignment. Code sets with a constant part are determined, step 100 , When codes are assigned, code sets with the constant part are used, step 102 ,

4A . 4B . 4C and 4D FIG. 13 are diagrams of preferred circuits for reducing complexity in the calculation of Λ _R. In 4A The number of codes processed by the two-level data detector is converted into a lookup table 62 is stored, and the zu _{R associated} with this code number is used. In 4B The number of codes processed by the two-level data detector is converted into a lookup table 64 and an unscaled Λ _R is created. The unscaled Λ _R , for example, by a multiplier 66 scaled by P _means whereby Λ _R is generated.

In 4C becomes the code matrix C or the code identifier in a lookup table 68 stored. Using the lookup table 68 Λ _{R is} determined. In 4D becomes the code matrix C or the code identifier in a lookup table 70 entered, whereby an unscaled Λ _{R is} generated.

The unscaled Λ _R , for example, by a multiplier 72 scaled by P _means whereby Λ _R is generated.

Claims (29)

Method for recovering symbols from Signals received in a shared spectrum the method comprising: Processing codes of in the shared spectrum received signals using a Block-Fourier transform FT and generating a block diagonal code matrix; Appreciate one Channel response of the received signals; Expand and Modify the channel response to produce a block-circulatory matrix, and Perform a Block FT and generating a block diagonal channel response matrix; Combine the block diagonal code matrix and the block diagonal channel response matrix; Scan the received signals; Processing the received signals using the combined block diagonal code matrix and the block diagonal channel response matrix with a Cholesky algorithm; Perform a block inverses FT for a result of the Cholesky algorithm to match spread symbols produce; and Despread the splayed symbols to symbols recover the received signals. The method of claim 1, wherein the Cholesky algorithm determining a Cholesky factor and performing a forward and backward substitution includes. The method of claim 1, wherein combining the block diagonal code matrix and the block diagonal channel response matrix adding a factor for the noise fluctuation multiplied by a unit matrix comprises. The method of claim 1, wherein the block diagonal Code matrix by multiplying a code matrix by a complex one conjugates transpose the code matrix and perform a Block FT of a multiplication result is generated. The method of claim 1, wherein the block diagonal Code matrix by entering a number of codes of interest is generated in a lookup table. The method of claim 1, wherein the block diagonal Code matrix by entering a number of codes of interest into a lookup table and scale out of the lookup table resulting block diagonal matrix with a medium power level is produced. The method of claim 1, wherein the block diagonal Code matrix by entering code IDs of the received signals is generated in a lookup table. The method of claim 1, wherein the block diagonal Code matrix by entering code IDs of the received signals into a lookup table and scale out of the lookup table resulting block diagonal matrix with a medium power level is produced. The method of claim 1, wherein the block diagonal Code matrix by inputting codes of the received signals into one Lookup table is generated. The method of claim 1, wherein the block diagonal Code matrix by inputting codes of the received signals into one Lookup table and scale one out of the lookup table resulting block diagonal matrix with a medium power level is produced. A data estimator for use in recovering symbols from signals received in a shared spectrum, the data estimator comprising: means ( 34 ) for processing codes of the signals received in the shared spectrum using a block Fourier transform FT and generating a block diagonal code matrix (Λ _R ); Means for estimating a channel response (H) of the received signals; Facility ( 36 . 38 ) for expanding and modifying the channel response to generate a block circular matrix (H _C ), and for performing a block FT and generating a block diagonal channel response matrix (Λ _H ); Facility ( 40 . 42 . 44 . 46 ) for combining the block diagonal code matrix (Λ _R ) and the block diagonal channel response matrix (Λ _H ); Means for sampling the received signals; Facility ( 48 . 20 . 22 . 24 . 58 ) for processing the received signals using the combined block diagonal code matrix and the block diagonal channel response matrix with a Cholesky algorithm; Facility ( 60 ) for performing a block inverse FT for a result of the Cholesky algorithm to generate spread symbols; and means for despreading the spread symbols to recover symbols of the received signals. The data estimator of claim 11, wherein the Cholesky algorithm comprises determining a Cholesky factor and performing a forward ( 22 ) and backward substitution ( 24 ). A data estimator according to claim 11, wherein the device ( 40 . 42 . 44 . 46 ) for combining the block diagonal code matrix (Λ _R ) and the block diagonal channel response matrix (Λ _H ) comprises means for adding ( 46 ) of a factor for the noise fluctuation multiplied by a unit matrix. A data estimator according to claim 11, wherein the device ( 30 . 32 . 34 ) for processing codes and generating the block diagonal code matrix means ( 32 ) for multiplying a code matrix (C) by a complex conjugate transpose of the code matrix (C _H ) and a device ( 34 ) for performing a block FT of a multiplication result to produce the block diagonal code matrix. The data estimator of claim 11, wherein the block diagonal code matrix (Λ _R ) is generated by inputting a number of codes of interest into a lookup table. The data estimator of claim 11, wherein the block-diagonal code matrix (Λ _R ) is generated by inputting a number of codes of interest into a look-up table and scaling a mid-level block diagonal matrix resulting from the look-up table. The data estimator of claim 11, wherein the block diagonal code matrix (Λ _R ) is generated by inputting code identifiers of the received signals into a lookup table. The data estimator of claim 11, wherein the block diagonal code matrix (Λ _R ) is generated by inputting coded identifiers of the received signals into a lookup table and scaling a block diagonal medium power level matrix resulting from the lookup table. A data estimator according to claim 11, wherein the block diagonal code matrix (Λ _R ) is generated by inputting codes of the received signals into a look-up table. A data estimator according to claim 11, wherein the block diagonal code matrix (Λ _R ) is generated by inputting codes of the received signals into a lookup table and scaling a block diagonal medium power level matrix resulting from the lookup table. The data estimator according to one of the claims 15-20, further comprising the lookup table. The data estimator of claim 11, wherein: the device ( 34 ) for processing codes of the signals received in the shared spectrum is a block Fourier transform device FT; the means for estimating a channel response (H) of the received signals is a channel estimator; the facility ( 36 . 38 ) for expanding and modifying the channel response, an expansion and modification device ( 36 ); the facility ( 40 . 42 . 44 . 46 ) for combining the block diagonal code matrix and the block diagonal channel response matrix is a circuit for combining; the means for sampling the received signals is a scanning device; the facility ( 48 . 20 . 22 . 24 . 58 ) for processing the received signals using the combined block diagonal code matrix and the block diagonal channel response matrix with a Cholesky algorithm, a Cholesky decomposition apparatus ( 48 ) and forward ( 22 ) and backward ( 24 ) Has substitution devices; the facility ( 60 ) for performing a block inverse FT for a result of the Cholesky algorithm to generate spread symbols, an inverse block FT device for performing a block inverse FT for an output of the backward substitution device ( 24 ) is to generate spread symbols; and means for despreading the splayed symbols to reproduce symbols of the received signals win, is a desaster. The data estimator according to claim 22, wherein the circuit for combining comprises two multipliers having. The data estimator of claim 22, further comprising: a hermetic device ( 30 ); and a multiplier ( 32 ) for multiplying a code matrix by a complex conjugate transpose of the code matrix. The data estimator according to claim 22, further comprising a look-up table and a multiplier wherein the block diagonal code matrix is input by inputting a Number of codes of interest in the lookup table and multiply a diagonal block matrix resulting from the lookup table is generated at a medium power level. The data estimator according to claim 22, further comprising a look-up table and a multiplier wherein the block diagonal code matrix is input by typing Codekennun gene of the received signals in the lookup table and Multiplying a diagonal resulting from the look-up table Block matrix is generated with a medium power level. The data estimator according to claim 22, further comprising a look-up table and a multiplier wherein the block diagonal code matrix is input by typing Codes the received signals into the lookup table and multiply a diagonal block matrix resulting from the lookup table is generated at a medium power level. Wireless Transceiver Unit (WTRU) that supports the The data estimator according to one of the claims 11-27. A base station comprising the data estimator of any of claims 11-27.